WP 2013-03 Revealed Preference and the Strength/Weight Hypothesis
ABSTRACT. The strength/weight hypothesis is the claim that people are more persuaded by extreme evidence (high strength) than by reliable evidence (high weight), even when these two dimensions of evidence are diagnostically equivalent according to Bayes Rule. We investigate whether this hypothesis is supported when tested in a revealed preference experiment. We confirm the hypothesis. Although earlier researchers using hypothetical, stated preference methods have found a pattern of overconfidence for high strength/low weight information, and underconfidence for low strength/high weight information, we find that underconfidence is the main direction of the bias. Moreover, the strength/weight effect in our revealed preference experiment was much smaller than in earlier stated preference studies.